In this paper, a novel competitive swarm optimizer (NCSO) is presented for large-scale global optimization (LSGO) problems. The algorithm is basically motivated by the particle swarm optimizer (PSO) and competitive swarm optimizer (CSO) algorithms. Unlike PSO, CSO neither recalls the personal best position nor global best position to update the elements. In CSO, a pairwise competition tool was presented, where the element that fails the competition are updated by learning from the winner and the winner particles are just delivered to the succeeding generation. The suggested algorithm informs the winner element by an added novel scheme to increase the solution superiority. The algorithm has been accomplished on high-dimensional CEC2008 benchmark problems and sampling-based image matting problem. The experimental outcomes have revealed improved performance for the projected NCSO than the CSO and several metaheuristic algorithms. © 2020, Springer Nature Singapore Pte Ltd.